| 64 | 0 | 758 |
| 下载次数 | 被引频次 | 阅读次数 |
针对电子数据取证与案件侦查中移动应用行为分析的技术需求，基于低频电磁辐射特征，对手机端运行的移动应用行为模式进行了识别与分析。提出了一种非接触式的移动应用行为识别方法，该方法通过采集手机运行特定应用程序时产生的低频电磁辐射信号，利用时频域分析技术将其转换为特征样本，随后构建并输入Spectrogram Transformer深度学习模型进行特征提取与分类识别。实验测试结果表明，所构建的模型在应用级分类任务中的最高准确率达到99.07%，在更细粒度的具体操作行为分类任务中准确率达到95.61%。研究结果验证了利用低频电磁辐射进行移动应用行为识别的可行性，该技术路线能够在电子数据取证的证据补强机制、非接触式侦查以及辅助验证设备操作人主观意图等方面提供客观的技术佐证。
Abstract：To address the technical requirements for mobile application behavior analysis in digital forensics and criminal investigation， this study identifies and analyzes behavior patterns of mobile applications running on smartphones based on low-frequency electromagnetic radiation characteristics， A non-contact mobile application behavior recognition method is proposed. This method collects low-frequency electromagnetic radiation signals generated by mobile phones running specific applications， using time-frequency domain analysis to convert them into feature samples， and subsequently employs a constructed Spectrogram Transformer deep learning model for feature extraction and classification， Experimental results demonstrate that the proposed model achieves a maximum accuracy of 99.07% in application-level classification tasks and achieves 95.61% in fine-grained specific operation behavior classification tasks. These findings validate the feasibility of using low-frequency electromagnetic radiation for mobile application behavior recognition. Furthermore， this technical approach offers objective technical corroboration for evidence reinforcement mechanisms in digital forensics， non-contact investigations， and the auxiliary verification of the device operator's subjective intent.
[1] Liu Y， Wang X， Qu B， et al. ATVITSC: A Novel Encrypted Traffic Classification Method Based on Deep Learning[J]. IEEE Transactions on Information Forensics and Security， 2024， 19: 9374-9389.
[2] 蒋鹏飞， 魏松杰. 基于深度森林与CWGAN-GP的移动应用网络行为分类与评估[J]. 计 算机科学， 2020， 47(01): 287-292.
[3] Kusuma R S， Putra M D P. Android Malware Threats: A Strengthened Reverse Engineering Approach to Forensic Analysis[J]. JISKA (Jurnal Informatika Sunan Kalijaga)， 2025，10(1): 122-138.
[4] Syeda D Z， Asghar M N. Dynamic Malware Classification and Api Categorisation of Windows Portable Executable Files Using Machine Learning[J]. Applied Sciences， 2024， 14(3): 1015.
[5] Sayakkara A， Le-Khac N A. Forensic Insights from Smartphones through Electromagnetic Side-Channel Analysis[J]. IEEE Access， 2021， 9: 111198-111209.
[6] Oberhansl F， Schink M， Kabakci N J， et al. Breaking ECDSA with Electromagnetic Side-Channel Attacks: Challenges and Practicality on Modern Smartphones[J]. arXiv preprint arXiv:2512.07292， 2025.
[7] Navanesan L， Le-Khac N A， Scanlon M， et al. Ensuring Cross-Device Portability of Electromagnetic Side-Channel Analysis for Digital Forensics[J]. Forensic Science International: Digital Investigation， 2024， 48: 301684.
[8] 魏忠诚， 董延虎， 陈虹宇， 等. 基于WiFi的多用户行为识别技术[J]. 无线电工程， 2025， 55(09): 1894-1906.
[9] Misek J， Jakus J， Hamza Sladicekova K， et al. Extremely Low Frequency Magnetic Fields Emitted by Cell Phones[J]. Frontiers in Physics， 2023， 11: 1094921.
[10] Ahmed S A A， Awais M， Wang W， et al. ASiT: Local-Global Audio Spectrogram Vision Transformer for Event Classification[J]. IEEE/ACM Transactions on Audio， Speech， and Language Processing， 2024， 32: 3684-3693.
[11] Xia C， Wang X， Lv F， et al. Vit-comer: Vision Transformer with Convolutional Multi-Scale Feature Interaction for Dense Predictions[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2024: 5493-5502.
[12] Han K， Wang Y， Chen H， et al. A Survey on Vision Transformer[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence， 2022， 45(1): 87-110.
[13] 徐童心， 黄俊. 基于CNN-Transformer的欺骗语音检测[J]. 无线电工程， 2024， 54(05): 1091-1098.
[14] Khanaghavalle G R， Anitha R. A Vision Transformer-Based Approach for COPD Detection Using Hybrid Time-Frequency Audio Features[J]. Signal， Image and Video Processing， 2026， 20(3): 122.
[15] Vaswani A， Shazeer N， Parmar N， et al. Attention is All You Need[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems. 2017: 6000-6010.
[16] Smietanka L， Maka T. Interpreting Convolutional Layers in DNN Model Based on Time-Frequency Representation of Emotional Speech[J]. Journal of Artificial Intelligence and Soft Computing Research， 2023， 14: 5-23.
[17] Saritha B， Laskar M A， Kirupakaran A M， et al. Deep learning-Based End-to-End Speaker Identification Using Time–Frequency Representation of Speech Signal[J]. Circuits， Systems， and Signal Processing， 2024， 43(3): 1839-1861.
[18] Ji X， Cheng Y， Xu W， et al. No Seeing is Also Believing: Electromagnetic-Emission-Based Application Guessing Attacks Via Smartphones[J]. IEEE Transactions on Mobile Computing， 2021， 22(2): 1095-1109.
[19] Ding D， Li Y， Zhang Y， et al. Sniffing the Application Usage Information with the Leakage Current of Laptops[J]. IEEE Transactions on Mobile Computing， 2026， 25(1): 934-947.
基本信息:
中图分类号:D918;TP18;TN98
引用信息:
[1]时百辰.基于低频电磁辐射的移动应用行为识别方法[J].无线电工程().
2026-05-22
2026-05-22
2026-05-22